RGB-D visual positioning system and method for complex structure

Abstract

The invention discloses an RGB-D visual positioning system and method for a complex structure. Normalization processing is carried out on a target image and a template image to realize data dimensionreduction so as to improve the calculation efficiency. Similarity comparison is performed after normalization, templates with the highest scores in k groups (k being an integer greater than 1) are selected, then, evaluation grouping is carried out according to viewpoint positions of the templates so as to cope with the situation of rotationally symmetric objects, finely, fine adjustment is performed on a target point cloud in a small angle range, finally, a best matching angle is found out, and a six-coordinate pose of the objects can be obtained through reverse operation. The method has the beneficial effect that the RGB-D positioning method has preferable robustness and also has broader adaptability for the shapes and poses of the objects through the innovation of an algorithm under thecondition of without increasing the cost.

Description

technical field [0001] The invention relates to the field of machine vision, in particular to an RGB-D visual positioning system and method for complex structures. Background technique [0002] See figure 1 , Traditional 3D vision often uses monocular cameras and structured light, or uses binocular and multi-eye cameras to obtain depth information through image calculations; or uses RGB-D cameras to directly obtain color images and depth images. Usually, it is necessary to perform feature extraction on the template image, calibrate the pose of the template, and then perform feature matching and calculation between the image of the object to be tested and the template, and finally obtain the limited pose of the object to be tested. The disadvantages are: 1) The calibration operation is complicated, and calibration failure is prone to occur, and the environment has a great impact on the use effect; in some poses, if the feature points are covered, it will cause positioning fa...

AI Technical Summary

Problems solved by technology

The disadvantages are: 1) The calibration operation is complicated, and calibration failure is prone to occur, and the environment has a great impact on the use effect; in some poses, if the feature points are covered, it will cause positioning failure; 2) The recognition and positioning of complex objects are accurate 3) Only suitable for limited pose determination of simple objects in a stable environment

Images